gopher-glide

README

Gopher-Glide (gg) 🚀

Run your IDE `.http` files under heavy concurrent load. Zero scripting required.

---

🤷‍♂️ Why not just use wrk, k6, or JMeter?

If you want to run a quick concurrency test on an endpoint, you usually have to write a custom JavaScript file for k6, a Python script for Locust, or learn a heavy configuration language.

Gopher-Glide (gg) is an API traffic simulator that lets you skip the boilerplate. It directly executes standard IDE .http REST Client files that you probably already have sitting in your workspace.

Feature	Gopher-Glide (gg)	k6 / Locust	wrk / hey / vegeta
Scripting	None (Reads .http natively)	JavaScript / Python	None (CLI flags only)
Traffic Control	30-FPS Interactive TUI (Arrow keys)	Requires configs	Fixed concurrency only
CI/CD Assertions	Semantic JSON Diffing (gg snap)	Pass/Fail Thresholds	Raw latencies only
Built-in Profiles	Yes (--profile flash-sale)	Requires scripting	No
IDE Integration	Native JetBrains Plugin	External scripts	External tools
Performance	30k+ RPS per core (Actor Model)	Medium-High	Extremely High

👉 Read the Full Documentation

---

⚡ 30-Second Quickstart

1. Install via Homebrew

brew install shyam-s00/tap/gg

2. Point it at any .http file

Create a simple api.http file (or use an existing one from your IDE):

GET http://localhost:8080/health
Accept: application/json

3. Run the simulation

Use a built-in profile to automatically shape the traffic (e.g., ramping up, sustaining, and cooling down):

gg --hive-engine --profile flash-sale --http-file api.http

You will instantly drop into the newly polished, butter-smooth 30-FPS Interactive Chaos TUI where you can monitor latencies and adjust the load in real-time without stutter or lag.

---

🛠 Feature Deep-Dives

🐝 The Hive Engine (Zero-Allocation Architecture)

The core of gg is built on a pure-Go lock-free Actor Model. By isolating connections into ultra-lightweight Goroutines and tracking metrics via sharded lock-free atomics, gg operates at 0 allocs/op on the hot path. This means you can comfortably push over 30,000+ RPS on a standard developer machine without garbage-collection pauses destroying your latency percentiles.

⚔️ Benchmark: At 30,000 RPS target load, gg consumes 40% less memory (1.42 GB vs 2.38 GB) and delivers 3x more successful requests than k6. gg's dynamic Adaptive Backpressure acts as an intelligent edge proxy—gracefully throttling excess traffic when the target server breaks to ensure maximum goodput, instead of blindly hammering the network into a deadlock. Read the full benchmark breakdown to see how gg safely simulates massive traffic.

🔬 Semantic Regression Gates (gg snap)

Don't just test if your API is slow—test if it's broken. By passing the --snap flag, Gopher-Glide infers your API's JSON schema in real-time. You can then use gg snap assert in your CI/CD pipeline to automatically break the build if a pull request spikes latency or alters a JSON payload contract.

🔌 JetBrains IDE Integration

Gopher Glide features an official JetBrains plugin that brings load testing directly into your IDE. Run profiles directly from your .http files via the gutter icon, and explore semantic diffs using the Snap UI Tool Window.

🔄 Stateful Journeys & Variable Extraction

No more writing custom JavaScript to test multi-step workflows. gg natively understands JetBrains HTTP variable extraction syntax. You can authenticate, extract a token using # @gg-export token = jsonpath: $.data.token, and instantly inject it into the next request using {{token}}.

---

Planned Features

• Chaos / Fault Injection — Simulate poor network conditions (3G packet loss, latency jitter) to see how APIs handle degradation.
• Distributed Simulation Mesh — Run gg worker nodes across multiple servers/regions, controlled by a central gg master UI.

License

MIT